Bug 35994

Summary: HP LP2465 screen is cutting in and out after upgrade to linux 2.6.28
Product: xorg Reporter: Tyson Whitehead <twhitehead>
Component: Driver/RadeonAssignee: xf86-video-ati maintainers <xorg-driver-ati>
Status: RESOLVED FIXED QA Contact: Xorg Project Team <xorg-team>
Severity: normal    
Priority: medium    
Version: 7.5 (2009.10)   
Hardware: x86-64 (AMD64)   
OS: Linux (All)   
Whiteboard:
i915 platform: i915 features:
Attachments:
Description Flags
dmesg from working patched 2.6.36 kernel
none
dmesg from non-working 2.6.38 kernel
none
xorg log from working patched 2.6.36 kernel
none
xorg log from non-working 2.6.38 kernel
none
xrandr verbose from working patched 2.6.36 kernel
none
xrandr verbose from non-working 2.6.38 kernel
none
avivo regs from working patched 2.6.36 kernel
none
avivo regs from non-working 2.6.38 kernel
none
possible fix none

Description Tyson Whitehead 2011-04-05 09:47:26 UTC
I have a HP/Compaq 8501w with a "ATI Technologies Inc M76 [Radeon Mobility
HD2600 Series]" graphics card.  It has a HDMI port on the side of it which I use to plug into a DVI monitor at home and work via a HDMI to DVI cable.

I was running linux 2.6.28 patched with

- bdd91b2b and fb939dfc (to fix 31339), and
- a0ae5864 (to fix 18564).

I upgraded to 2.6.38 and, while my monitor at home continues to work okay, my monitor at work now has flickering black lines and it cuts in and out to a black screen.  I did diffs of the kernel logs, xorg logs, and xrandr --verbose outputs, but nothing seemed to indicate any significant difference.

Finally, I

- checkout out the 2.6.38 kernel from git,
- reverted to bdd91b2b (the first version that had all the above patches), and
- discovered that it also had these problems, 

so I

- did a bisect on any changes that touched the radeon directory between that and 2.6.26 (cherry picking bdd91b2b, fb939dfc, and a0ae5864 as required), and
- this pointed to 48dfaaeb (drm/radeon/kms: remove new pll algo).

I then did some verification of this by

- attempting to revert 48dfaaeb on top of 2.6.38.

However, this failed as it was not clean and I was not sure of my corrections, so I instead

- reverted it on top of bdd91b2b (the first version that had all the above patches)

which produced a driver that worked fine with both monitors.

I presume this then means my monitor at work is having an issue with the pll settings the old new pll algorithm is generating.  I don't know what information would be helpful for resolving this though, so I'll just attach

- the kernel logs,
- the xorg logs, and
- the xrandr --verbose logs

for my original patched and working 2.6.36 kernel and the non-working 2.6.38 (although I couldn't see any significant differences in these).  Please let me know if there is any other information I can provide.

Thanks!  -Tyson
Comment 1 Tyson Whitehead 2011-04-05 09:48:38 UTC
Created attachment 45276 [details] [review]
dmesg from working patched 2.6.36 kernel
Comment 2 Tyson Whitehead 2011-04-05 09:49:05 UTC
Created attachment 45277 [details] [review]
dmesg from non-working 2.6.38 kernel
Comment 3 Tyson Whitehead 2011-04-05 09:49:37 UTC
Created attachment 45278 [details] [review]
xorg log from working patched 2.6.36 kernel
Comment 4 Tyson Whitehead 2011-04-05 09:50:10 UTC
Created attachment 45279 [details] [review]
xorg log from non-working 2.6.38 kernel
Comment 5 Tyson Whitehead 2011-04-05 09:50:42 UTC
Created attachment 45280 [details] [review]
xrandr verbose from working patched 2.6.36 kernel
Comment 6 Tyson Whitehead 2011-04-05 09:51:13 UTC
Created attachment 45281 [details] [review]
xrandr verbose from non-working 2.6.38 kernel
Comment 7 Alex Deucher 2011-04-05 10:38:54 UTC
Can you use the radeonreg tool and dump the regs with the working and non-working kernels?
http://cgit.freedesktop.org/~airlied/radeontool/
./radeonreg regs avivo
Comment 8 Tyson Whitehead 2011-04-05 11:02:16 UTC
Created attachment 45304 [details]
avivo regs from working patched 2.6.36 kernel
Comment 9 Tyson Whitehead 2011-04-05 11:02:58 UTC
Created attachment 45305 [details]
avivo regs from non-working 2.6.38 kernel
Comment 10 Alex Deucher 2011-04-06 08:56:10 UTC
Created attachment 45348 [details] [review]
possible fix

Does this patch help?
Comment 11 Tyson Whitehead 2011-04-06 16:25:53 UTC
Hi Alex,

I applied your patch on top of the stock 2.6.38 driver, compiled it up, and it works great with both my monitor at work and at home.

Thanks!  -Tyson

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.